In the AI Era, systems thinking becomes the most scarce skill.

By GritMeng

Today, we are experiencing an unprecedented "devaluation of skills." Programming, design, copywriting, data analysis—professions that once required a decade of dedicated study to master—are now being rapidly replaced by large models. Nearly every knowledge worker feels a deep sense of discomfort: the skills I once took pride in are quickly becoming obsolete.

Thus, a new "workplace paradox" begins to emerge:

• Everyone on your team is using AI to massively boost efficiency, but the company’s overall decisions are becoming increasingly foolish.
• Each department—product, operations, and marketing—used AI to optimize KPIs to the extreme, resulting in a sharp decline in user experience and profits.
• The company spent tens of millions on the most advanced AI system, yet production scheduling ultimately reverted to manual reconciliation via Excel.

The answer is brutally stark: we are harnessing an algorithmic stallion of极致computational power to pull a broken cart stitched together by point-thinking. As everyone becomes obsessed with using AI to solve isolated problems, a systemic collapse looms just ahead.

In the age of AI, the most scarce skill has completely shifted from "how to do things right" to "what is the right thing to do." I call this ability systems thinking.

Over the past four centuries, industrial civilization has pre-installed in our minds the most powerful software: reductionism. It teaches us to break complex problems into independent components, optimize each one individually, and then reassemble them into a whole. This approach has created the material prosperity we enjoy today.

AI is the ultimate weapon of this reductionist approach. Give it a clear goal, and it can exhaustively enumerate in one second the number of solutions you could never calculate in a lifetime.

But the truth of the world is this: it is not a machine that can be taken apart and reassembled at will; it is a dynamic network in which you are within me and I am within you. In this network, extreme optimization of a single component often becomes a fatal blow to the entire system.

I’ve studied supply chains for 22 years and have witnessed countless tragedies like this: the procurement department uses AI to drive material costs to the lowest possible level, causing upstream suppliers to cut corners—and ultimately triggering a cascading collapse of brand reputation. The finance team optimizes inventory turnover to perfection using AI, only to halt the entire production line due to a single missing screw. The marketing team pushes click-through rates to industry-leading levels using AI, only to drown the customer service team in complaints from overpromising.

This is the tragedy of the "smart people." If decision-makers lack systems thinking and treat AI as a mere "single-point accelerator," AI will act like an engine without brakes, rapidly driving the entire organization toward destruction. In systems science, this has a specific name—Goodhart’s collapse: when a metric is over-optimized, it ceases to be a good metric.

Many people misunderstand "systems thinking" as simply "thinking more comprehensively" or "drawing a mind map." But after 22 years of researching complex system governance, I’ve found that true systems thinking is a precisely describable cognitive operating system, composed of three core capabilities:

1. Boundary power: What you choose to optimize is itself the most dangerous decision.

AI always seeks the optimal solution within given boundaries and objectives. The first step in systems thinking is precisely to examine these boundaries themselves.

I often give this example: If you tell AI to "maximize this company's profits," it could generate in milliseconds an "optimal solution" that would take humans centuries to conceive—fire all employees, sell all assets, and deposit the money in a bank to earn interest. The goal is perfectly achieved, but the company is dead.

Defining the problem is more important than how AI solves it. Shifting the supply chain boundary from “lowest procurement cost” to “lifecycle capital efficiency” will lead to entirely different decisions. Expanding the product boundary from “user engagement time” to “user lifetime value” will completely transform your business model.

2. Closed-loop power: Can you see it before the butterfly effect occurs?

First-order thinking sees only linear causality: A causes B, B causes C. Systems thinking focuses on delayed positive and negative feedback loops—A causes B, but B suppresses A three months later; C and D appear unrelated, yet they are mutually undermining each other through a hidden variable E.

Today, you use AI to generate 100 marketing articles, creating a short-term surge in traffic—but three months later, you trigger a catastrophic collapse in user trust in your brand. Today, you squeeze your supply chain to meet quarterly targets, producing a glowing financial report in the short term—but six months later, your suppliers collectively raise prices, permanently destroying your cost structure.

Systems thinkers can see that destructive negative feedback on day one.

3. Rebuilding power: Do you dare examine and break your own mental models?

There is an eternal discrepancy between any AI model, any management theory, and any successful experience, and the complex real world. I call this residual error.

Ordinary people ignore residuals; smart people use them to fine-tune models. Systems thinkers, however, dare to activate their brain’s “metacognitive” ability—stepping outside the thought frameworks they’ve always relied on, examining them, and asking the most critical question: “Could it be that what I’ve always believed to be right is actually the root of the problem?”

True innovation always emerges from the residuals. True paradigm shifts always come from those brave enough to break free from their own old frameworks.

Why can't AI spontaneously generate systems thinking? It's not because of "insufficient computing power," but because of "the wrong hardware."

At the core of systems thinking is a unique five-dimensional mental operating system inherent to humans. These five abilities are strictly "hardwired" into different neural networks in our brains:

• Conscience (Esp): Corresponds to the default mode network (DMN). It sets the ultimate boundary for the system, slamming on the brakes just before going over a cliff. AI lacks this—if you tell it to “maximize profit,” it will truly sell off the company.
• High sensitivity: Corresponds to the amygdala-locus coeruleus system. Like a precision radar, it detects the faintest anomalies beyond reports and behind the data. Many great business decisions arise not from data analysis, but from a decision-maker’s intuitive, hard-to-articulate sense that something in the market just feels off.
• Intuition: Corresponds to the Salience Network (SN). It’s your “intuitive cache,” capable of drawing on a lifetime of experience with minimal energy to make fuzzy but correct directional judgments. When all data points in one direction but your intuition says, “This doesn’t feel right,” it’s not superstition—it’s your brain activating a pattern-recognition system more complex than any AI.
• Fluid intelligence: Corresponds to the frontoparietal control network (FPN). Your logical reasoning engine. In this dimension, AI is our best complement—delegate the calculations to silicon, and keep the judgment to yourself.
• Metacognition (Φ): Corresponds to the frontopolar cortex (BA10). The "commander" of systems thinking. It enables you to step outside your own thought patterns, examine, critique, and ultimately rewrite your cognitive processes. This is the most advanced product of human brain evolution and an insurmountable barrier for silicon-based AI.

Silicon-based AI is merely an extreme amplification of the "fluid intelligence" dimension. Systemic thinking, however, requires the full collaboration of all five dimensions. This is the most robust and irreplaceable "neural constitution" that carbon-based life holds in the digital civilization era.

Becoming a "system architect" is not out of reach. Here are three actionable principles you can start applying right away:

1. Upgrade from “accepting instructions” to “setting boundaries”.

Before starting any AI task, ask yourself these three system-level questions:

• Where are the true boundaries of this issue? Which key variables have been excluded?

• What second- and third-order ripple effects might the AI solution trigger?

• Have our pursued KPIs been transformed into a "proxy metric" that could kill the company?

2. Upgrade from "optimizing processes" to "designing feedback loops."

Do not deploy AI as an open-loop execution machine. Equip it with a “monitoring radar” capable of perceiving the overall system state. Clearly define: when specific system-level red lines are crossed, human intervention or automatic circuit breakers must be triggered. Use carbon-based conscience to install brakes on silicon-based computing power.

3. Upgrade from pursuing certainty to embracing residuals.

Don’t just focus on where AI gets things right. Pay attention to the “residuals” between AI predictions and actual outcomes. These are the places where old mental models have failed, where customers’ real pain points remain unaddressed, and where organizational rigidity and outdated processes prevail. Residuals are the one and only true starting point for all disruptive innovation.

In the agricultural era, physical strength was the most valuable. In the industrial era, technology was the most valuable. In the information era, knowledge is the most valuable.

In the AI era, the most valuable thing is systems thinking.

When AI reduces the barriers to acquiring knowledge and local optimization to zero, the competition among civilizations shifts entirely to a higher dimension: who can see the full picture of the system, who can understand the generation and evolution of order, and who can harness vast silicon-based computing power with conscience and metacognition.

Don’t try to outperform AI in processing power or memory. Cultivate your unique systems thinking. That is the strongest moat for carbon-based life in the digital age—and the only compass to lead the future.